The Day Three Major FMT × Immunotherapy Trials Landed Together in Nature Medicine — Reshaping Cancer Immunotherapy with the Gut Microbiome | Making Cancer Immunotherapy Work with FMT, Vol. 1

Key Takeaways

• In April 2026, Nature Medicine Vol. 32 No. 4 published three clinical trials of fecal microbiota transplantation (FMT) combined with immune checkpoint inhibitors (ICIs) back-to-back: FMT-LUMINate (phase 2, NSCLC + melanoma), PERFORM (phase 1, mRCC), and TACITO (randomized phase 2, mRCC).
• FMT-LUMINate hit an objective response rate (ORR) of 80% in NSCLC against historical 39−46%; TACITO produced a median progression-free survival of 24.0 vs 9.0 months (hazard ratio 0.50, P=0.035) — both breaking through the long-standing ceiling that “only half of patients benefit from ICIs.”
• The deeper significance lies not in the numbers but in the shared scientific finding: responders were not “patients who acquired the donor’s beneficial bacteria” but “patients who lost their own deleterious bacteria.” FMT-LUMINate proved this causally in mice.
• This article is Volume 1 of the series “Making Cancer Immunotherapy Work with FMT.” We survey the three trials in one sweep and explain why their simultaneous publication marks a clinical inflection point — the foundation for the deeper mechanistic and commercial discussions in Volumes 2 and 3.

Introduction — The “Publishing Event” of Three Papers in One Issue

In science, there are rare moments when several major papers on the same topic appear in the same issue of a journal — not because the editors curated a special feature, but because independent groups, working in parallel, have all reached publication-ready evidence at the same time. April 2026’s Nature Medicine Vol. 32 No. 4 was such a moment.

Pages 1316–1324: the TACITO trial from Gianluca Ianiro and colleagues in Italy. Pages 1325–1336: the PERFORM trial from Saman Maleki Vareki and colleagues in Canada. Pages 1337–1350: the FMT-LUMINate trial from Arielle Elkrief and colleagues, also in Canada. Three FMT × ICI clinical trials, lined up consecutively in the same issue.

FMT — fecal microbiota transplantation — transfers stool from a healthy donor (or a donor with specific clinical attributes, such as a complete responder to ICIs) into a recipient’s gut. It is already standard care for recurrent Clostridioides difficile infection. In cancer immunotherapy, the field’s interest in FMT began around 2015, when the hypothesis that gut microbes shape ICI efficacy moved from mouse models into human cohorts and, eventually, from rescuing refractory cases to first-line combinations. About a decade of cumulative evidence brought us here.

The April 2026 trifecta signals that this accumulated evidence has crossed a threshold. This article surveys the three trials in one view, extracts three converging findings, and explains why this publication moment is decisive. Detailed mechanistic dissection follows in Volume 2, and the commercial-and-regulatory outlook in Volume 3.

Main Body

1. Half-Effective: The Clinical Ceiling of ICIs

Over the past decade, immune checkpoint inhibitors (ICIs) have transformed solid-tumor oncology. PD-1/PD-L1 inhibitors and CTLA-4 inhibitors — used as monotherapy or in combination — are now standard or major options across nearly every major solid tumor: lung, melanoma, renal, gastric, urothelial, head-and-neck, and hepatocellular cancers.

And yet the everyday clinical reality is captured bluntly: “only half of patients benefit.”

• NSCLC (PD-L1 high) on pembrolizumab monotherapy: ORR 39−46%.
• Melanoma on nivolumab + ipilimumab: ORR 50−58%.
• Metastatic RCC (mRCC) on pembrolizumab + axitinib: ORR around 60%, with most progressing within 16 months.

The remaining patients either fail to respond from the outset (primary resistance) or develop resistance after an initial response (acquired resistance). Why does the same drug work dramatically in some patients and not at all in others? One answer that biology has put forward — increasingly forcefully — is: “It may be the microbes living in your gut.”

2. The Microbiome Hypothesis — A Lineage from 2015 to 2026

In 2015, L. Zitvogel and colleagues in France published a landmark Science paper showing that the antitumor effect of CTLA-4 blockade in mice depended on specific Bacteroides species in the gut (Vétizou et al., 2015). The same year, T. Gajewski’s group in Chicago showed that Bifidobacterium amplifies the effect of PD-L1 blockade.

In 2018, three near-simultaneous Science papers (Routy / Gopalakrishnan / Matson) demonstrated in human cohorts that gut microbiome composition predicts ICI response. Antibiotic exposure shortly before ICI was repeatedly linked to worse outcomes.

In 2021, two Science papers (Baruch et al.; Davar et al.) provided the first human evidence that FMT could overcome ICI resistance in melanoma — by transplanting stool from complete responders into patients refractory to anti-PD-1.

In 2023, Routy and colleagues published the MIMIC trial in Nature Medicine, showing the safety and activity of healthy-donor FMT in front-line melanoma combined with anti-PD-1.

And then April 2026: three trials at once. From “rescue of refractory cases” to “first-line combination,” from “melanoma alone” to “multiple cancer types,” and from “exploratory data” to “randomized controlled comparison” — all three frontiers of FMT × ICI research advanced together.

3. The Three Trials at a Glance

The three trials, compressed into one table:

4. FMT-LUMINate — 80% ORR in NSCLC, and a Striking Mechanistic Discovery

FMT-LUMINate enrolled 20 treatment-naive NSCLC patients (PD-L1 ≥ 50%) and 20 treatment-naive cutaneous melanoma patients. Each received a single dose of healthy-donor FMT via oral capsules, followed by ICI initiation.

The headline result: NSCLC ORR of 80% (16/20), dwarfing the historical 39–46% benchmark for pembrolizumab monotherapy and exceeding the pre-set efficacy threshold of 64%. One-year PFS 65%, one-year OS 100%, median duration of response 8.7 months. In melanoma, ORR was 75%, including 4 complete responses, against historical 50–58%.

However, the melanoma cohort showed an earlier-than-expected onset of dual-ICI toxicity: 60% grade 3 AEs, one grade 4, and a striking 15% incidence of myocarditis (against <1% in the general literature). This concentrated in recipients of one specific donor (donor 5, rich in Prevotella spp.) — a finding we will return to in Volume 3.

The most striking scientific finding of FMT-LUMINate is this: what distinguished responders from non-responders was not engraftment of donor-derived “good” bacteria, but loss of the patient’s own pre-existing “harmful” bacteria. Shotgun metagenomic profiling showed no significant difference in donor-strain engraftment between responders and non-responders. Instead, responders selectively lost Enterocloster citroniae, Clostridium innocuum, Enterocloster lavalensis, and other taxa previously implicated as “bad actors” in ICI resistance. The team then ran a reverse experiment in antibiotic-pretreated mice: they reintroduced precisely those species that had been lost in human responders — and the antitumor effect of FMT was abolished. Loss of deleterious species, not gain of beneficial ones, mediated the therapeutic effect of FMT. Causality was established.

5. PERFORM — A Curious “Responders Have Fewer Side Effects” Pattern in mRCC

PERFORM enrolled 20 treatment-naive mRCC patients (intermediate or poor IMDC risk) and gave them a healthy-donor encapsulated FMT product (LND101, developed at Lawson Health Research Institute) before ICI-based therapy. 80% (16/20) received anti-PD-1 + anti-CTLA-4; the rest received anti-PD-1 + a VEGF inhibitor.

The primary endpoint was safety. Grade 3+ irAEs occurred in 50% (10/20), within the expected range. No serious FMT-related toxicities, no grade 4–5 events. Secondary outcomes: ORR 50% (9/18 evaluable), CR 11% (2/18), median PFS 11.15 months, median OS 36 months.

Clinically, the most intriguing finding was that only 1 of 9 responders (11%) experienced a grade 3+ irAE, while 8 of 9 non-responders (89%) did (χ²=8.0, P=0.005). The conventional rule-of-thumb in ICI therapy — “the more side effects, the more efficacy” — was sharply inverted here.

Mechanistically, patients with higher α-diversity and durable engraftment of donor-derived anti-inflammatory functional pathways (short-chain fatty acid producers in particular) had both fewer toxicities and higher response rates. Conversely, patients who developed grade 3+ irAEs showed enrichment of Segatella copri (formerly Prevotella copri clade A) — a candidate toxicity-driving taxon shared with FMT-LUMINate. We discuss this in Volume 3.

6. TACITO — The First Real RCT in the Field, and a “Near-Miss with a Twist”

TACITO randomized 50 treatment-naive mRCC patients 1:1 to receive donor FMT (d-FMT, sourced from two complete ICI responders) or placebo FMT (p-FMT), administered alongside pembrolizumab + axitinib three times over 6 months (initial colonoscopy, then capsules at weeks 12 and 24). It is the first proper RCT in FMT × ICI.

The primary endpoint was the 12-month PFS rate. Result: d-FMT 70% (16/23) vs p-FMT 41% (9/22), P=0.053 — just shy of the prespecified threshold. A naive headline would say “primary endpoint not met.”

But the secondary endpoint of median PFS told a different story: d-FMT 24.0 months vs p-FMT 9.0 months, hazard ratio 0.50, P=0.035. In absolute terms, time to progression was 2.7 times longer. The per-protocol analysis confirmed a 12-month PFS of 71% vs 38% (P=0.046). In the IMDC intermediate / poor risk subgroup, the gap widened further: 18.8 vs 5.1 months (P=0.033).

The authors argue that “12-month PFS is a single-time-point snapshot that may underestimate FMT effects, which take time to manifest.” The microbiome analysis was telling: the donor-strain engraftment rate (DoSER) did not predict 12-month PFS. Instead, qualitative changes — acquisition of specific donor strains, or loss of specific recipient strains — drove outcomes, mirroring the FMT-LUMINate finding.

7. Three Convergent Findings Across the Three Trials

Read side-by-side, the three independent studies converge on three findings.

Convergent finding ① upends the implicit design philosophy of FMT to date: “deliver beneficial bacteria as reliably as possible.” With causality demonstrated in mice, the next generation of interventions should be designed around selective elimination of deleterious taxa.

Convergent finding ② has direct operational consequences for donor screening. Excluding Prevotella-rich donors (especially Segatella copri) from dual-ICI combination trials has already been adopted in CanBiome2 (NCT06623461, planned n=128).

Convergent finding ③ — most starkly visible in PERFORM — suggests FMT may also reshape the toxicity profile of ICI, with major implications for patient quality of life. Patients who responded but had to discontinue ICI for toxicity may find a new therapeutic window via FMT-style microbiome modulation.

8. The Commercial Reality — Who Will Build This, and When?

The clinical signal is now clear. The next question is operational: who will turn this into therapy?

The companies developing, manufacturing, and distributing FMT-class therapeutics have consolidated over the past few years. Seres Therapeutics (US) holds approval for VOWST/SER-109, an oral live biotherapeutic product (LBP) for C. difficile. Vedanta Biosciences (US, Flagship Pioneering) is developing LBPs (e.g., VE800) for cancer-immunotherapy combinations. Finch Therapeutics (US) has scaled back its cancer pipeline. In Canada, Lawson Health Research Institute / Western Ontario continues to advance LND101 (the encapsulated FMT used in PERFORM). The successor to FMT-LUMINate — CanBiome2 — will retest healthy-donor FMT in a 128-patient RCT.

The larger trend is a shift away from whole-stool FMT toward defined consortia (rationally designed LBPs): Faecalibacterium prausnitzii-based products such as Exeliom Biosciences’ EXL01 (France); multiple Akkermansia programs; and the early stirring of “selective phage-based removal of harmful taxa” as a complementary modality. Volume 3 will explore these.

9. Three Caveats — How to Read These Results Soberly

To avoid overheating, three limitations:

First, small sample sizes. FMT-LUMINate (NSCLC 20, melanoma 20), PERFORM (20), TACITO (FAS 45). All are phase 1–2 with sample sizes well below the 3,000+ typically required for registration-level RCTs. The signals are real, but these are not approval-grade evidence.

Second, donor dependence. As FMT-LUMINate’s toxicity concentration in donor 5 illustrates, results depend strongly on which donor is selected. Reproducible FMT medicine requires standardized donor screening — composition, Prevotella ratio, infectious-disease screening, functional enzyme profiles.

Third, backbone-regimen dependence. Segatella copri drives toxicity under ipi/nivo but is harmless under anti-PD-1 monotherapy. The same stool can behave differently depending on the partnered ICI. Discussions of FMT efficacy must always be paired with the backbone ICI.

Summary

• In April 2026, Nature Medicine Vol. 32 No. 4 published three FMT × ICI clinical trials (FMT-LUMINate, PERFORM, TACITO) back-to-back, generating cross-tumor evidence (NSCLC, melanoma, mRCC) all at once.
• FMT-LUMINate produced 80% ORR in NSCLC; TACITO produced a median PFS of 24.0 vs 9.0 months (HR 0.50, P=0.035) — both breaking the long-standing ICI ceiling.
• The deepest convergent finding is that response is driven by loss of the patient’s own deleterious bacteria, not engraftment of the donor’s beneficial bacteria — a fundamental redirection of FMT’s design philosophy. Causality was demonstrated in mice in FMT-LUMINate.
• Simultaneously, Segatella copri has emerged as a toxicity-driving taxon under dual ICI, and donor-screening practices are already being revised in successor trials.
• Commercialization is shifting from whole-stool FMT toward rationally designed consortia LBPs, with multiple full-scale RCTs (CanBiome2 and others) reading out in 2026–27.
• Limitations (small N, donor dependence, regimen dependence) duly noted, the April 2026 issue marks a watershed at which the field moved from “promising hypothesis” to “clinical application.”

My Thoughts and Outlook

The deepest implication of these three trials is not the numbers but the relocation of the lever of intervention. A decade ago, we hypothesized: “deliver good microbes and immunotherapy will work.” The 2026 data ask us to rewrite that as: “remove bad microbes and immunotherapy will work.” Having watched this field as a researcher, an investor, and an executive, I see this as a deep difference between an “additive” and a “subtractive” drug-development philosophy.
For Japanese clinical practice and industry, three near-term implications stand out. First, FMT and next-generation LBPs are emerging, within 2–3 years, as a real option for ICI-resistant or ICI-intolerant patients in clinical-trial settings. Second, donor screening — particularly the evaluation of Prevotella abundance — urgently needs standardization, and there is a participation opportunity here. Third, in the rational-consortium LBP space, players who can bridge Japanese-origin F. prausnitzii or Akkermansia-related strains into the global pipeline remain few in number.
Over the next three to five years, this field will evolve from “FMT, the blunt instrument” to “precision microbial cocktails.” For clinicians, this expands the therapeutic repertoire. For patients, it improves the trade-off between efficacy and toxicity. For industry, it opens a new revenue axis. Cancer immunotherapy’s microbiome chapter has only just begun.

Coming Next

Volume 2 will dissect the central mechanistic finding shared across the three trials — that the loss of deleterious bacteria, not the engraftment of beneficial ones, is what makes FMT work — focusing on FMT-LUMINate’s analyses. Why are Enterocloster citroniae, Clostridium innocuum, and Enterocloster lavalensis implicated in ICI resistance? What is the link to tryptophan metabolism and the kynurenine pathway? And what does the reverse-experiment in mice teach us about causality? We will explain, in plain language, the scientific case for redesigning FMT around subtraction rather than addition.

Edited by the Morningglorysciences team.

Related Articles

• Reading Breast Cancer Diagnosis with AI, Vol. 1: What MASAI Answered (already published)
• Making Cancer Immunotherapy Work with FMT, Vol. 2: Subtraction, Not Addition (forthcoming)
• Making Cancer Immunotherapy Work with FMT, Vol. 3: Donor Strategies and the Commercialization Crossroads (forthcoming)